DMARC has three possible policies, and the difference between them is not a nuance: it is the difference between observing a problem and solving it.
| Policy | What the receiver does | Protects? |
|---|---|---|
p=none | Delivers the forged email normally and sends you a report | No |
p=quarantine | Sends it to spam | Partial |
p=reject | Rejects it and it never arrives | Yes |
With p=none, someone can send a fake invoice today claiming to come from your company and it will land in your client's inbox exactly as if it were yours. You will receive a report telling you it happened. That is all.
Why so many stop at none
Because p=none is the right first step, and nobody takes the next one.
Observation mode exists for a good reason: before blocking, you need to know who legitimately sends in your name. It is almost always more of them than you remember — the invoicing platform, the campaign tool, the ticketing system, the ERP. If you switch on reject without having aligned them, your own invoices stop arriving.
So none gets set, the reports start arriving as unreadable XML, nobody processes them, and the project stays there. Months later the domain "has DMARC" on the security inventory and nobody notices it protects nothing.
What we found in Mexican brands
We looked up the DMARC policy published by ten large Mexican domains on 28 July 2026:
| Domain | Policy | Situation |
|---|---|---|
| telmex.com | p=none | no protection |
| grupobimbo.com | p=none | no protection |
| cemex.com | p=none | no protection |
| aeromexico.com | p=quarantine | to spam |
| banorte.com | p=reject | protected |
| bbva.mx | p=reject | protected |
| mercadolibre.com.mx | p=reject | protected |
| liverpool.com.mx | p=reject | protected |
| hp.com | p=reject | protected |
| soriana.com.mx | p=reject | protected |
Six out of ten went all the way to reject, which is the goal. But three are still in observation mode. All ten publish a reporting address, so all of them started the project; three did not finish it.
Note that both banks in the sample are on reject. Where impersonation costs money immediately and obviously, the work gets finished.
Google and Microsoft are already pushing it
Since 2024, Google and Yahoo have required DMARC from anyone sending email in volume to their users, and Microsoft announced equivalent requirements. The floor they ask for is p=none, so meeting the requirement and being protected remain different things — but the direction of travel is clear, and it heads towards demanding policies with teeth.
What policy does your domain have?
Our checker tells you whether your DMARC protects or merely observes, along with the state of your SPF and DKIM.
How to leave none without breaking anything
- Read the reports. They arrive as raw XML; you need a tool that aggregates them or someone to interpret them. Without this step the rest are done blind.
- Draw up the list of who sends for you. The reports reveal it, including the services nobody remembered.
- Align each one, with DKIM signed using your domain, not the provider's. This is where the real work usually is.
- Move up to
quarantinewith a partialpct. You can apply the policy to a fraction of the traffic (pct=25) and raise it as you confirm nothing legitimate is being caught. - Finish on
reject. Without this step, everything before it was instrumentation, not protection.
_dmarc.<domain> over DNS-over-HTTPS and read the p tag. These are public lookups and anyone can repeat them. The data is from 28 July 2026 and changes as soon as each company modifies its record.Keep reading
- SPF, DKIM and DMARC explained without jargon What each one does and why the three together decide whether your email arrives.
- The 10-lookup SPF limit that breaks your email without warning Two large Mexican brands are at the exact limit.